Discussion Strong ID for new and old systems

[Sleep Runner]

I have a question about Journal and system identification.

I'm currently working on collecting data from few CMDRs journals into database (to do some research), but some systems has the same name. Well... okay, it happens. So I can't count on system name anymore.
In 3.0 update we got SystemAddress (64bit ID). But old journal records doesn't have it.

Now I'm about using position as unique (string) ID. Like " 50.12345 : -50.12345 : 99.12345 ".

So the question is - May it happened in past versions that some system has changed it's coordinates? Or coordinates precision has been changed?
For example in December 2017 i see "Epsilon Indi" StarPos is 3.125 / -8.875 / 7.125, only .3 precision. It still the same on EDSM. But after 3.0 I see some systems has .5 precision, like System: "Luyten's Star" 6.5625 / 2.34375 / -10.25 .
May it happen that new records with this system come with 5 digit precision instead of old 3?

Guys from other database projects not talking to much. Docs also quiet about it. Google absolutely useless.

I'm cold, scared and confused.

 

[Sleep Runner]

Well... I decided to use checksum from name and position

 

[AnthorNet]

Before, the .3 was rounded.
Every set of coordinates can only stick on a 1/32 grid.

It's basically the same coordinates, before we were multiplying them to 32, rounding, then dividing again by 32 to have the right set.